Research map / Detection and segmentation / CNN features
Object Instance Annotation With Deep Extreme Level Set Evolution
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Use DELSE instead of DEXTR | Interactive instance segmentation · Cityscapes-Stretching | mIoU: 81.59 -> 83.35 | Helped |
| Use DELSE instead of Polygon-RNN++ | Interactive instance segmentation · Cityscapes-Hard | mIoU: 71.38 -> 73.84 | Helped |
| Use DELSE instead of DEXTR | Interactive instance segmentation · PASCAL | mean IoU: 90.5 -> 90.5 | No clear effect |
| Use DELSE instead of DEXTR | Interactive instance segmentation · PASCAL + SBD | mean IoU: 91.5 -> 91.3 | Hurt |
| Add the motion term and evolve the level set (ablation) | Level Set Regression on DAVIS 2016 · DAVIS 2016 | J mean: 81.7 -> 84.0 | Helped |
| Add the modulation term for selective curvature regularization (ablation) | DELSE with motion term on DAVIS 2016 · DAVIS 2016 | J mean: 84.0 -> 84.8 | Helped |
| Add skip features (ablation) | DELSE with motion and modulation terms on DAVIS 2016 · DAVIS 2016 | J mean: 84.8 -> 85.6 | Helped |
| Remove extreme-point input (ablation) | DELSE on DAVIS 2016 · DAVIS 2016 | J mean: 85.6 -> 83.9 | Hurt |
| Add ±5% noise to extreme-point input (ablation) | DELSE on DAVIS 2016 · DAVIS 2016 | J mean: 85.6 -> 85.1 | Hurt |
| Add ±10% noise to extreme-point input (ablation) | DELSE on DAVIS 2016 · DAVIS 2016 | J mean: 85.6 -> 84.9 | Hurt |
| Add motion editing corrections (ablation) | DELSE* trained on 10 of 16 Cityscapes cities · Cityscapes | mIoU: 82.45 -> 84.73 (1 click); 85.97 (2); 86.83 (3) | Helped |
| Add extreme-point correction clicks (ablation) | DELSE* trained on 10 of 16 Cityscapes cities · Cityscapes | mIoU: 82.45 -> 83.60 (1 click); 84.49 (2); 84.94 (3) | Helped |
Closest papers by meaning
- Large-Scale Interactive Object Segmentation With Human Annotators (CVPR 2019 · similarity 0.88)
- Training-Free Dataset Pruning for Instance Segmentation (ICLR 2025 · similarity 0.86)
- The Devil is in the Object Boundary: Towards Annotation-free Instance Segmentation using Foundation Models (ICLR 2024 · similarity 0.86)
- A Generative Approach at the Instance-Level for Image Segmentation Under Limited Training Data Conditions (Student Abstract) (AAAI 2025 · similarity 0.86)
- GrabS: Generative Embodied Agent for 3D Object Segmentation without Scene Supervision (ICLR 2025 · similarity 0.84)
- Extreme Point Supervised Instance Segmentation (CVPR 2024 · similarity 0.84)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.